Best-Practices Marketing Database Content has been a consistent theme of this monthly e-Letter. A key characteristic of Best-Practices Marketing Database Content is "completeness." Completeness takes three forms:
- Every customer and inquirer "event" is captured, even if its marketing value is not immediately apparent. This includes all orders, items, post-demand transactions and promotion history. Ideally, if available, customer service history is included.
- The descriptors ("fields") that describe every event are maintained at the lowest feasible level. This is important because, although you can always aggregate, you can never disaggregate.
- Archiving of older data generally should be avoided.
Savvy direct marketers understand that completeness results in full historical "views" for accurate, efficiently-executed analytics. For example, by not rolling transaction information off a Marketing Database after "say "36 months, all customers with 37+ month recency in a specific merchandise category will be properly represented in a product affinity or item-specific cluster analysis. Therefore, the conclusions derived from the analysis will be accurate representations of reality.
Many direct marketers do not realize that, when appropriate storage and access technologies are employed, completeness can actually be less expensive. This is because it eliminates the need for certain data-intensive processes such as archiving (along with the inevitable un-archiving), and outside-the-database workarounds to support customer-side reporting and analytical efforts such as:
- Matchback processing for response analysis.
- The creation of "time-0 views" for data mining projects such as predictive models, cohort analysis and over-time comparisons.
Some third-party contracts for Marketing Database hosting are driven by per-thousand run charges. This results in visible cost penalties for completeness because more records means higher charges. However, it is important to remember the offsetting, less obvious savings to be had from completeness; that is, the processing that is not done, as described in the previous paragraph.
